Fix: prevent useless pathfinder run for blocked vehicles

[glx22]

Fix #7670 and supersede #7822.

Motivation / Problem

As found in #7670, a road vehicle blocked by another vehicle may enter in an "loop" of intensive pathfinder calls.
This happens because starting tile for path search is not cached and as the vehicle did not move since previous call, the cached path is discarded and a new one has to be determined.

Description

This solution is similar to the one in #7822, but a lot less invasive and does not require a savegame bump.
When a road vehicle is blocked by another one, just push its current location and the direction to follow found by pathfinder on top of the existing cache.
That way on next run it will just retrieve the heading instead of recalculating it.
While in the area I noticed the speed of the blocked vehicle was not updated, so I fixed it.

Limitations

I only tested using the save from #7670 (comment) but I don't see how it could cause issues.

Checklist for review

Some things are not automated, and forgotten often. This list is a reminder for the reviewers.

• The bug fix is important enough to be backported? (label: 'backport requested')
• This PR affects the save game format? (label 'savegame upgrade')
• This PR affects the GS/AI API? (label 'needs review: Script API')
  • ai_changelog.hpp, gs_changelog.hpp need updating.
  • The compatibility wrappers (compat_*.nut) need updating.
• This PR affects the NewGRF API? (label 'needs review: NewGRF')
  • newgrf_debug_data.h may need updating.
  • PR must be added to API tracker

[LordAro]

Fine by me

[SamuXarick]

I wondered if the pathfinder prevention checks should be also added to VETS_CANNOT_ENTER, but I wasn't able find a situation where HasBit(r, VETS_CANNOT_ENTER) is true for testing that.